Re: poly: Why Oldies Stations?

Carl Feynman writes:
> Here's a model that I believe explains this process. Suppose there are a
> number of possible items in a certain category to choose from (items of
> clothing, dishes, types of music). Any particular item will benefit me a
> certain amount each time I indulge in it. The utility of each item is
> fixed, but I can't know its utility until I try it. The utilities of the
> items are drawn from a fixed distribution. Every time I go to indulge
> myself in an item of a certain category, I face the choice of picking one
> that I don't know about, or picking the best known choice so far. Let's
> assume that I have only a known finite number of times left to enjoy this
> particular category. Under these assumptions, the rational thing to do is
> to stop experimenting after some fraction of one's lifespan, and then stick
> with what one knows is best.
>
> Does this suffice to explain the phenomenon?

This reminds me of the gas station problem. (It has a proper name in
stat./prob., but I can't remember what it is.)

You are driving down a highway past a series of gas stations. You
cannot turn around, but you would like to pick a gas station with
a price reasonably close to the lowest one. Statistics can tell you
how many gas stations to pass before you know enough to pick a
reasonably priced gas station.

This model probably works for other things, like (I hate to put it
this way) picking an optimal mate. One can't "go back", so (perhaps)
one has a period of experimentation to judge what the parameters of the
"possibility space" is.

Perry
Received on Sat Apr 4 00:14:12 1998